Zhihe GrowthAn SSOT governs facts reused in websites, structured data, cases, and sales materials. Without a field-level master record, editors may update body text but not Schema, revise case metrics while leaving old homepage summaries, or turn regional prices and lead times into universal commitments. Conflicting pages increase the risk of AI misinterpretation.
Each master-table row should answer seven questions
| Fields | Questions to answer | Example: |
|---|---|---|
claim_id, entity |
Which fact about which entity is this? | Legal entity or stage-specific case metric |
| Value & Unit | How should the value be interpreted? | 17%, not 17 percentage points |
| Conditions and metric definition | For which question set, period, and platform is it valid? | Stage-test scope; not a cross-project comparison |
| Source | Who approved it, and what permits disclosure? | Official web case page/customer authorization |
| Validity and Version | When does it take effect? | Change date, old version archive |
| Visible Location | Which pages use it? | Text, tables, JSON-LD, downloads |
| Permissions | Public, internal, awaiting verification, or conflicting | Do not publish without approval |
The current public figures for Zhihe Growth's named GEO cases are stage-specific official-site citation rates of 17% for SuperPDR and 15% for ELEREIN. They concern different projects and should not be ranked directly. Reconcile differing figures in older reports against metric definitions, samples, and dates before reuse. Do not claim delivery results for other partner brands where collaboration scope is unspecified. See Criteria for disclosure of cases and Boundaries of cooperation.
Map from main table to page
Map the legal company name to visible About-page text and Organization markup; service names to service pages and Service markup; and article titles and update dates to visible content and TechArticle. JSON-LD represents public, verifiable facts, not hidden internal plans. Google's Structured data policy requires markup to represent visible content. Adding undisclosed cases to Schema to give AI more information is inappropriate.
For changes, verify the fact, update affected pages, check machine-readable markup, then record the version. A change in customer disclosure permission requires checking guides, citation cards, image alt text, Schema, and caches, not just deleting a name from one case page. Legal identity, patent status, customer names, outcome figures, prices, terms, and certifications need stronger review than ordinary definitions.
An example of conflict management
If a service page says Shenzhen headquarters while an old article says Hangzhou headquarters, do not choose from memory. Check the current approved company profile and registration records to establish the Shenzhen entity's relationship to the Hangzhou branch or operations center. Give them separate SSOT fields, then check site search, footers, Schema, and third-party citations for consistency. If evidence is insufficient, flag the fact for review instead of concealing the conflict.
Zhihe Growth can use this table as a publication gate, delivering field differences, evidence gaps, and a revision list. Private customer ledgers remain private. Related reading: brand entity disambiguation and the role and limits of Schema.
Make updates an auditable state machine
A high-risk fact progresses through awaiting verification → verified → approved for disclosure → published → review required or withdrawn. Editors may propose changes but must not bypass source and approval checks. Builds should reject private or unverified values in JSON-LD. When a source fails, flag it for review and decide whether to preserve a historical fact, replace the source, or withdraw the claim. After removing text, check sitemaps, PDFs, captions, and caches for old versions.
Automation can use claim_id to generate an impact list instead of blindly replacing text. A stage-specific case citation rate may appear in details, overviews, service descriptions, and machine-readable catalogs; review its context on each page. Do not drop denominators, platform scope, or question-set conditions during reuse. Log what changed on which pages without exposing internal approvers or private test details, giving customers traceable deliverables while keeping public pages concise.